Software Development

After having spent over a (completely fruitless) month on fiddling with this without having an actual Force Touch Trackpad to test with, having gotten one recently, it still was insanely difficult for me to finally figure out how to receive Force Touch events from an NSTableView. But I finally did it. Here’s how.

Apple Keynote - MacBook Force TouchImage Credit: Apple, Inc.

The Premise

In Yoink, the usual way to select all files (pressing ⌘-A like you would in almost any application) is not a viable option. Yoink is written to not have key focus, as it would otherwise interfere with work in other applications too much (believe me, I’ve tried).
So I had to come up with a different way to do it. What I’ve done up until now is to allow the user to hold down the option (⌥) key on the keyboard and double-clicking onto any file in Yoink to select all files.

When Force Touch was introduced, I thought that would be a very nice alternative to option-double-clicking. And it is. Once I got it working, anyway.

The Issue

What I thought would be the easiest way to get Force Touch events from an NSTableView was to override – (void)pressureChangeWithEvent:(NSEvent *)theEvent.
Only that it doesn’t get called as it would normally be called: continuously, after – (void)mouseDown:(NSEvent *)theEvent, with every change of pressure on the trackpad.

No, in an NSView-based NSTableView (I haven’t tested cell-based tableViews), -pressureChangeWithEvent: gets called at -mouseDown: with a pressure of 0.0, and then just “dies”, not getting called again until the next -mouseDown:.
That’s interesting. And there’s more interesting-ness going on.

Once I override the tableView’s -mouseDown:, -pressureChangeWithEvent: suddenly gets called continuously, as you’d expect. So something apparently happens inside the tableView’s -mouseDown: implementation that sort of prevents -pressureChangeWithEvent: from getting called.

Finding A Solution

Knowing that the table view’s default implementation of -mouseDown: somehow interferes with the easy way of using -pressureChangeWithEvent:, my first attempt at a solution was to override -mouseDown: and re-implement its functionality myself, because then I would receive -pressureChangeWithEvent: properly. Then it dawned on me what I was in for. Dragging something out of the tableView (something kind of essential in Yoink) suddenly didn’t work anymore. I’d have to re-implement the selection mechanism (selecting with a mouse click, de-selecting with the command key pressed, adding an additional selection with either the shift- or the command key, etc.)
That alone would have taken more time than the entire feature of having force-click-to-select-all is worth.

Next, I thought NSGestureRecognizer could work. I created a subclass of it that overrides -pressureChangeWithEvent: and checked to see if it was called. But it was the same thing all over again. Adding it to the tableView, the NSTableRowView, the NSTableCellView and the underlying contentView of NSWindow, trying to figure out if they maybe swallowed the method call yielded no result or any change. It was all the same. It got called at the first -mouseDown: call, but then stopped.

Next I tried a local NSEvent monitor thinking it might do the trick, and it works, but only when the window’s contentView is force clicked, not the tableView. A global monitor doesn’t seem to work with pressure events.

So the only way that seemed to get me anywhere at all was to override NSTableView’s -mouseDown:. And that’s what I ended up doing. At first. After a day’s work, though, it still didn’t yield any results what-so-ever. The amount of time I’ve wasted on this is astounding, but apparently, when you keep going and really want to get somewhere, you eventually will.

The Solution

In OS X 10.10 Yosemite, Apple introduced a new NSEvent-tracking API in NSWindow: – (void)trackEventsMatchingMask:(NSEventMask)mask timeout:(NSTimeInterval)timeout mode:(NSString *)mode handler:(void(^)(NSEvent *event, BOOL *stop))trackingHandler that lets you, in a tracking loop, monitor for events matching the mask you provide, in this case, NSEventMaskPressure.

I actually had played around with this before, coming recommended from Markus Müller (@fafner on twitter), developer of the marvellous Mac and iOS app MindNode, who, by the way, in order for me to be able to test this stuff before I got my own Force Touch-able Trackpad, kindly invited me to their office to use his MacBook – thank you, Markus.
However, I didn’t get very far, as I didn’t know how to use the API correctly. Calling it from – (void)awakeFromNib, for example, isn’t the best idea, as it a) doesn’t appear to be able to track anything and b) locks up the app. It has to be called from inside a tracking loop (like -mouseDown: or -pressureChangeWithEvent:).

Again, after a lot of time wasted getting nowhere, this is the solution I came up with:

Source Code for how to receive Force Touch events from an NSTableView

I decided to imitate NSTableView’s -action and -doubleAction methods for handling receiving force touch events in a target.
NSTableView can be set up with a -target that should receive the selectors set in -action-doubleAction, and with this, -forceTouchAction, as seen in the code above, uses the same approach – send the specified selector to the NSTableView’s -target.

This code starts tracking Pressure-, Drag- and Mouse Up events in -pressureChangeWithEvent:, and practically waits until the Pressure event’s stage 2 is reached, which is a force touch. Then it sends the -forceTouchAction to -target.

The important lesson that I learned here is that the events have to be forwarded to the window, as they’re intercepted. So calling [self.window postEvent:event atStart:NO]; is imperative.

I watch for NSLeftMouseUp and NSLeftMouseDragged because I need to be able to stop tracking at some point, and mouseUp and mouseDragged are the perfect moments to do that.

Sample Project

I’ve uploaded a quick sample project to my server which you can download here. It requires OS X 10.10.3 (because of the APIs used). I’ve only tested it On 10.11 El Capitan, though, so your mileage may vary.

I hope this sample code is useful to you and saves you all the headaches I’ve experienced (and a lot of time) figuring this stuff out. Maybe I’m not smart enough. Or maybe this is harder than it should be. I dream of a time where the Force Touch APIs on OS X get the loving treatment 3D Touch enjoys on iOS 😉

Read more

In the latest Xcode beta (currently, 7.1 beta 2), I noticed this in its Release Notes (login required) :

(…)Developers running on Yosemite will not be able to pair the Apple TV Remote with the tvOS Simulator Runtime. (…)

That implies that it’s somehow possible to pair the Apple TV Remote with a Mac running OS X 10.11 El Capitan – but how? There’s no explanation anywhere to be found.

Un-pair the Apple TV Remote

If you’ve already paired your Apple TV Remote with your Apple TV Developer Kit, you won’t be able to pair it with your Mac – you’ll have to un-pair it first.

Remote and interaction remote 2xOriginal Image Credit: Apple Inc., markup mine.

A response to this forum threadon the Apple Developer Forums lead me to how to do it: Hold down the Menu and Volume-Up keys on your Apple TV Remote for about 5 seconds, perhaps a little longer, and the Remote’s pairing will be reset.

Pair it with your Mac

Launch System Preferences and click on Bluetooth. In there, you’ll see something like this:

Screenshot of System Preferences 05 10 2015 17 37 23The Apple TV Remote in System Preferences / Bluetooth.

Click on Pair to pair the Apple TV Remote with your Mac. There should be a spinning wheel for a couple of seconds after which it should be paired.

Ready to Rock

Now the Xcode tvOS Simulator automatically recognizes the paired Apple TV Remote and should respond to anything you do on it. Enjoy 🙂

Update (October 6th, 2015) – Media Control?

Oliver Drobnik (@Cocoanetics on twitter) was interested in what kind of Bluetooth profiles the Apple Developer Kit Remote supports – for example, the Media Control profile to be able to adjust volume, play/pause, etc) – or if it can only be used in Xcode’s Simulator.
He suggested I use the free Bluetooth Scanner App LightBlue to see what profiles it supports.
According to the app and this Apple website that lists the service names, the profiles supported are Battery Service, Bond Management, Device Info and a custom Apple service – so no media control for now.
But thanks to Oliver, now we know.

Read more

[Note: This is a guest blog post written by Martin Nguyen (@iMaddin on twitter), an OS X- and iOS developer based in Austria, about his Mac app Gestimer]

Gestimer Icon

I don’t like to keep too many thoughts in my head at the same time. Also, my short-term memory isn’t the greatest. If I had to cook dinner and then went back to my Mac to continue whatever I was working on, I’d probably forget about my dinner.

That’s especially true while coding where I have to be completely focused. Other unrelated thoughts are too distracting when writing code. I feel comfortable knowing that an app will take care of such thoughts as its memory is most likely better than mine.

Not too long ago – while I was at university – I noticed that todo apps weren’t entirely fitting all my needs. Sometimes I had these little tasks like reminding me to make dinner (because I’d forget and be too busy or focused with other things) or that I had to leave for a lecture in half an hour. It was too much of a hassle to set alerts for those tasks that have a short lifespan. To me it just feels slow to enter numbers with a keyboard for inputing dates and times, especially without a num pad.

As the type person who sits at my Mac most of the time, I decided to make a Mac app to work around this problem. After a couple of months and failed attempts at such an app, the idea that later became Gestimer popped into my mind: “Wouldn’t it be great if I could just drag down from the menu bar to create a quick reminder?” This seemed so simple and fast. It was the perfect solution.

Development

With this idea in my mind and a few sketches in my notebook, it was time for the execution. This was back in 2012 when I had just begun learning to code. As every beginner knows, coding is frustrating. Everything can break easily and you have no clue why. Objective-C was also not the easiest thing to learn without any prior programming knowledge. Additionally, making Mac apps appeared to be much more difficult than making iOS apps.
I shelved the project. There was nothing that was anywhere close to Gestimer out there so it was impossible finding resources that would help me realize the app.

With iOS 7, Apple introduced UIKit Dynamics. It seemed like a fun way to make interactions so I played around with it. I noticed that UIKit Dynamics allowed me to easily do something close to what I had in mind. It was only available for iOS and not OS X but it still allowed me to produce a version of Gestimer, even if it wasn’t in the intended environment. Here is the iOS version which was available from 2013 to 2014:

Gestimer_iOS

Gestimer for iOS didn’t do too well but that’s okay. I didn’t do any marketing as I simply put it up on the App Store and I was already happy to have an app that I made and used every day. Creating the iOS version taught me a lot and as I learned from more projects over the last couple years, I felt comfortable enough to tackle the Mac version again. I was especially motivated by the announcement and release of Swift.

I made some good progress on Gestimer for the Mac during the summer of 2014. The idea never changed: drag & drop from the menu bar to create a reminder. I worked on the app on and off, whenever I felt like it, whenever I slept off the frustration and head-scratching of the previous day.

As there was no UIKit Dynamics for the Mac, it was a lot more difficult to imitate the iOS version and to get things behaving as intended. By late 2014 I had a rough but useable version. I was never in a rush to release Gestimer. In fact, I stopped working on it for a couple months after that. I was again content with having an app I made and used every day. It also gave me time to simply use the app and think about if it truly did the things well that I wanted it to do.

If you haven’t seen Gestimer yet, have a look at a short clip here.
https://vine.co/v/ehEwHbWKhUj/embed/simple

Marketing

Sometime around mid-April 2015 I started sharing a beta of Gestimer with people who I thought might be interested. Besides gathering feedback, it got me excited about the prospect of launching and hear what even more people will say about it. That’s where I decided to do some proper marketing. I won’t go into detail about this here as it would double the size of this post but I took marketing very seriously.

The basic outline of what I had:
– a short 10s clip of the interaction on the website
– collected email addresses for launch day
– made a YouTube video (currently over 50k views)
– a nice press kit
– localized Gestimer into 9 languages (now even more!)

Launch

There was little press coverage on launch day, but I had Gestimer up on Product Hunt thanks to Matthias who kindly sent me an invite. Someone had also posted the app to Hacker News. Both sites led to a lot of traffic. At the end of the day Gestimer was sitting on the 3rd spot on Product Hunt. To this date, the PH submission for the app has received over 400 upvotes. Twitter also helped a bit, though I believe people overestimate its effectiveness.

More press coverage followed and an increasing number of people spread the word. During launch week Gestimer reached the #1 spot in the Top Paid charts on the Mac App Store in multiple countries including the US, UK, and Germany. While sales numbers on the Mac App Store are small compared to the iOS App Store – even if you reach the top charts – it’s still been good. It’s nothing to complain about.

The reception for Gestimer has been overwhelming and I would have never imagined it becoming this successful. It makes me very proud to have done all the design, code, and marketing work on my own.

If you’d like, you can find out more about Gestimer here.

——
Martin Nguyen (@iMaddin on twitter) sometimes works on iOS and OS X apps and lives in Austria. While he enjoys coding, he is still trying to figure out the next step in his life.

Read more

It was a bug I’ve been carrying along for quite some time in Yoink. But I finally found the culprit: I’m looking at you, OS X sandbox.

Webarchive opened after bug occurredA webarchive created with Safari, after a security-scoped NSURL bookmark was created for it.

The Bug and its Detection

As it happens (sadly), not I discovered the bug, but a customer and user of my app Yoink encountered it. The reason being, I rarely handle .webarchive files (if at all) – webarchive files are created when saving a website in Safari, for example – but one of Yoink’s users, Christoph, appears to have to deal with them regularly.

The bug itself is described fairly easily. You have a .webarchive file you’d like to move using Yoink. So you drag it onto the app and then move it from Yoink to the actual destination. Business as usual.
Only that now, instead of opening the webarchive in your standard browser when double-clicked, above’s warning is shown.

There’s two baffling things about this warning:

  1. It looks like you’re trying to launch an application instead of just a file (“from an unidentified developer”)
  2. The creator of the file seems to have changed. Instead of Safari, it now says “BugReport-WebArchivesAndNSURLBookmarks” (the app I submitted to Apple to demonstrate this issue)

Highly concerning, to say the least. To some people, it might even look like something fishy is going on.

Hunting Down the Bug

Seeing as .webarchive files are binary property lists (thank you, Michael Tsai (@mjtsai on twitter) for the correction), I tried other property list files (for example, .plist files), but none exhibited the same behavior.

Reproducing it was fairly easy. There’s one pre-requisite for it to appear: that in System Preferences -> Security & Privacy -> Allow apps downloaded from, either ‘Mac App Store’ or ‘Mac App Store and identified developers’ be selected.
Otherwise, Gatekeeper isn’t active and doesn’t react to the issue.

System Preferences - Security and Privacy Settings

At first I thought it happened when moving the file out of Yoink, since that’s where Yoink actually does something to a file – it moves it from one location to another – who knows what goes on behind the scenes there.

Going through the move-code line by line, commenting out stuff I thought could cause this, made no difference what so ever. Create a new webarchive, move it using Yoink, and get the warning again. Rinse and repeat.
With stuff like this, I get annoyed easily, so my patience usually goes out the window after a couple of alterations to the code.

It was only after the billionth time that I considered it might happen when a file was added to Yoink, not moved from Yoink.

Going through the same process as before, I went through the code line by line, trying different things.

With a recent update to Yoink, the app knows when a file is renamed or deleted in Finder, using GCD (Grand Central Dispatch) and its dispatch sources.
I believed the issue lay there, but after commenting out all of that code as well, it became clear it wasn’t the culprit.

In a fit of anger and a severe feeling of incompetence, I randomly picked at the code (I know, highly professional. But I’m using version control – all is well, right?).

To my surprise, when I removed the code responsible for saving Yoinks files over relaunches of the app, the issue went away.

Security-Scoped NSURL Bookmarks

For a sandbox’ed app to keep a reference to a file added by the user beyond relaunches, it has to use what is called a security-scoped NSURL bookmark.

Sandbox entitlements necessary for security-scoped bookmarks

They can be used if the corresponding entitlement (see above) is added to the app’s sandbox entitlements file.

There are two ways a security-scoped bookmark can be created – either with read and write access, or read-access only (NSURLBookmarkCreationWithSecurityScope or it plus NSURLBookmarkCreationSecurityScopeAllowOnlyReadAccess).

I don’t know why this causes the issue, but changing the bookmark creation options from read and write access to read-access only fixed the issue (and moving the file is still possible with Yoink). It definitely looks like a sandbox bug to me.

Why should an app-internal bookmark (mind you, it’s only a reference to a file, not the file itself, created from an NSURL object) write to and change the webarchive’s file so that it a) can’t be opened anymore and b) shows the bookmark-creating-app as its creator?

Bug Reporting to Apple

I reported this bug to Apple (rdar://21765077) with an example project you can download here, if you’d like to see for yourself. Feel free to dupe the bug with Apple’s Bug Reporter Tool – I’d appreciate it 🙂

Now all that’s left to say is thank you for your patience, Christoph. You reported this issue at the beginning of March 2015 and had to wait until now for it to be fixed (and still a little longer because I have to submit the update to Apple for release on the Mac App Store as well). I really appreciate your continued feedback on the app!

Correction

Thanks to Michael Tsai for pointing out on his blog that .webarchive files aren’t bundles, as I falsely stated, but binary property lists.

Read more